Analysis and experimental exploration of a nanofiltration membrane system, in the context of potential urine treatment processes.

The flux continued to increase with increasing Transmembrane Pressure (TMP) between 800 and 1250 kPa for the XLE membrane, indicating that higher TMP conditions are usable without loss of energy efficiency. Fouling resulted in a negligible decrease in flux for the NF 270 membrane, a 15 % decrease in flux for the NF 90 and an 18 % decrease for the XLE membrane, all of which are within permissible limits.

C ONTEXT OF THE STUDY

P URPOSE OF THE STUDY

R ESEARCH O UTCOMES

S IGNIFICANCE OF THE STUDY

L IMITATIONS OF THE STUDY

I NTRODUCTION

U RINE B ACKGROUND

Due to the human element in the use of UDT, some cross-contamination of urine is to be expected. This means that some faecal matter may also be present in the urine stream and must also be removed before use as fertilizer.

Table 1 shows the variation in urine composition. This variance is due to many factors including diet, which has a significant effect on pH and nutrient composition, and health, along with the accompanying medication taken, of the source

T REATMENT O BJECTIVES

D ISINFECTION
V OLUME R EDUCTION
S TABILISATION
P HOSPHOROUS R ECOVERY
N ITROGEN R ECOVERY
N UTRIENT R EMOVAL
M ICRO - POLLUTANT R EMOVAL

Phosphorus is mainly used in the fertilizer industry and is obtained from the mining of phosphorite, the phosphate-rich mineral. An example of this would be the conversion of any nitrogen present in urine to N2, which can be safely released into the atmosphere.

T REATMENT P ROCESSES FOR U RINE D ISPOSAL OR B ENEFICIATION

E VAPORATION
M EMBRANE F ILTRATION
N ITROGEN AND A MMONIA R ECOVERY
O THER

This process results in almost 100% removal of the water in the urine and an easy-to-handle solid. This would mean that most of the water in the urine would have to be extracted beforehand.

Figure 3: Relative Comparison of Desalination Costs and energy requirements [20]

P ROCESS S ELECTION

T REATMENT P ROCESS A NALYSIS

D ESIGN C ONSTRAINTS

The treatment system will be installed in remote areas and will operate mostly autonomously and thus should be low maintenance and easily repairable. Thus, biological systems, which are food-specific and do not tolerate large flow changes, and systems that are technologically complex or new, should be avoided.

U RINE T REATMENT P ROCESS S ELECTION

The first approach, shown in Figure 6, would concentrate the urine by separating the water from the urine, possibly using a combination of forward osmosis and membrane distillation. The problem with this process stream would be the potential for significant contamination in the first stage, as all organics and salts would still be present. The second approach, shown in Figure 7, would reduce the fouling potential in the first stage by using a nanofiltration membrane, which is more resistant to fouling.

A final forward osmosis (FO) stage (or combination of forward osmosis and membrane distillation) separates the remaining salts from water.

N ANOFILTRATION

D EFINITION
S EPARATION M ECHANISM
M EMBRANE T YPES
F OULING
N ANOFILTRATION OF URINE

The solution-diffusion mechanism achieves separation based on the solubility and diffusivity of the compounds and the permeability of the membrane. A detailed overview of the rejection by the NF 270 membrane, for both synthetic and fresh urine, can be found in Figure 9. The effect of permeation flux and pH on the accuracy of the model remains to be investigated.

The literature shows that the repulsion of many of the ions shown varies considerably.

Figure 9: Fresh and synthetic urine rejections from Pronk et al. [22]

G AP A NALYSIS

Knowledge gaps in the use of nanofiltration for urine treatment are seen as the most important missing information as achieving the required salt separation is a key factor in all three proposed configurations. Therefore, the remainder of this work focuses on nanofiltration, specifically investigating salt rejections and water flux. As stated in section 2.6.1, the focus of the experimental section of the project was on nanofiltration.

The results of the experiments were then used to determine whether it was possible to retain most of the valuable minerals, including potassium, phosphorus, and nitrogen, while allowing sodium chloride to pass through the hydrolyzed urine.

R ESEARCH D ESIGN

The membranes used were 3 different DOW-Filmtec membranes, the NF270, NF90 and XLE membranes, as they are readily available and in widespread commercial use. Another reason to use DOW-Filmtec membranes is that the NF 270 membrane was used in the work of Pronk et al. 22], discussed in the literature review, which allowed a direct comparison of the results with the same membrane.

The membranes have MWCOs ranging from 100 to 400, with XLE having the smallest and NF 270 the largest, and are made of polyamide.

E XPERIMENTAL D ESIGN

N ANOFILTRATION E QUIPMENT

The pressure drop can be measured across the cells using 2 pressure gauges with an accuracy of Pressure drop across the cells and flow rate were controlled using a combination of valve opening in the return line and pump speed. The reflux valve was used for rapid changes in flow rates, and thus pressures, and also served as a means of reducing hydrostatic shock across the membranes during startup and shutdown.

Figure 12: High Pressure Cross-flow Membrane Laboratory Rig

E XPERIMENTAL P ROCEDURE

A NALYTICAL E QUIPMENT

The Spectroquant Nova 60 is a spectrophotometer which uses the absorbance of light from the solution to calculate the concentration of the ions present. The concentration of the substance will affect the intensity loss of the light passing through the solution. In general, an atomic emission spectrometer (AES) allows one to calculate the concentration of specific elements by measuring the intensity of light emitted at a wavelength specific to each element.

The pH of the solution was measured using pH strips, resulting in an accuracy of ±0.5.

D ATA A NALYSIS

T RANSPORT M ODEL
E FFECT OF M EMBRANE A REA
W ATER R ECOVERY
S EPARATION F ACTORS

The flow of solute from the feed to the permeate would decrease the solute concentration in the feed over time. Although this change is negligible for this system, the effect can be shown by the increase in surface area in the above model due to the small permeate flow compared to the inlet volume. A good distribution between NPK and Na does not help if this only happens at low water volumes, as this would mean that most of the nutrients would be carried into the retentate.

If the separation factor is equal to 1, it means that there is no difference in rejecting component 'i' or 'j'.

A NALYTICAL E QUIPMENT

S PECTROQUANT N OVA 60 O PERATION

The main objectives of this project, as mentioned in section 1.3, were to place membrane processes in the context of urine treatment, to identify the knowledge gaps that prevent the use of membrane systems for urine processing, and finally to explore critical knowledge gaps for the use of nanofiltration through experiments. The specific knowledge gaps identified in section 2.6.1 and detailed in section 3.1 can be summarized as follows: verifying salt rejections reported in the literature and determining whether it would be possible to capture the majority of the valuable retain minerals, including potassium. phosphorus and nitrogen, while allowing the permeation of the unwanted sodium chloride into the hydrolyzed urine. These samples were prepared with a dilution factor between 20 and 100, which ensures that the measurement ranges discussed are lower than the actual sample concentrations.

Accuracy can be found by first calculating the standard deviation of the repeated measurements of the same sample, then comparing the standard deviation with the detection range.

A NALYTICAL P RECISION

The accuracy for these readings was excellent for the samples with lower concentrations, below 10 mg/l, and quite good for higher concentrations, 10 to 100 mg/l. The easiest way to compare this error across the different concentration ranges was to use the standard deviation of the measurements and then take it as a percentage of the mean value for that element. The results, shown in Table 5, show that the measurements for the synthetic urine were less accurate than for the stored urine.

This makes sense as the preparation procedure for Nova 60 is more complex than the dilution required for Agilent 4100.

Table 5: Stored and Synthetic Urine Composition and Analytical Precision

F LUX R ESULTS

W ATER F LUX VERSUS T RANSMEMBRANE P RESSURE
S YNTHETIC AND S TORED U RINE F LUXES
S OLUTION F LUX VERSUS T RANSMEMBRANE P RESSURE
F OULING P OTENTIAL

The stored urine flow was expected to be lower because the pollution was expected to be higher than the pollution caused by the synthetic urine. This seemed the likely case because the synthetic was made from laboratory-grade ingredients, while the stored urine contained organic compounds and particles, but this is not the case. The lower flux for the synthetic urine may be due to the osmotic pressure difference between the synthetic and stored urine.

Another unexpected result was the higher flux for stored urine through the XLE membrane than through the NF 90 membrane, while the synthetic flux of urine and water follows the opposite trend.

Figure 16: Average Synthetic and Stored urine flux for 3 membranes, at a TMP of 800 kPa

R EJECTION R ESULTS

R EJECTIONS BY M EMBRANE T YPE

R EJECTION VS T RANSMEMBRANE P RESSURE

C OMPARISON TO L ITERATURE

N ANOFILTRATION U SAGE

T RANSPORT M ODEL

E FFECT OF INCREASING MEMBRANE AREA ON FEED CONCENTRATION
E FFECT OF INCREASING MEMBRANE AREA ON PERMEATE CONCENTRATION
W ATER RECOVERY
S EPARATION FACTORS

Using the transport model described in Section 3.3.1, the area was increased by multiplying the initial area by 10 and 50, the recovery increased to 10% and 52%, respectively, for the NF90 membrane with the synthetic urine. The flux would be adequate for RTTC purposes and corresponded to literature values. Boller, “Nanofiltration for separation of pharmaceuticals from nutrients in source-separated urine,” Water Res., vol.

Modeling the separation performance of nanofiltration membranes for the mixed salt solution with Mg2+ and Ca2+,” J.

Figure 24: Concentration profile of species on permeate side

U RINE C OMPOSITION

F RESH U RINE

A NALYSIS E QUIPMENT

S PECTROQUANT N OVA 60 SOP S

The color of the measuring solution remains stable for at least 60 minutes after the end of the reaction time mentioned above. In such cases it is advisable to perform a plausibility check of the measurement results by diluting the sample. The color of the measuring solution remains stable for 30 minutes after the reaction time mentioned above.

The color of the measuring solution remains stable for 30 minutes after the end of the reaction time mentioned above.

R ESULT S HEETS

W ATER F LUX

S YNTHETIC U RINE

S TORED U RINE

Urea Hydrolysis [10]

Ammonium equilibrium [10]

In addition to the energy required, some of the volatile substances in the urine, such as ammonia, can be lost if the urine is not properly pre-treated. This low pressure is caused by the compression and subsequent condensation of the water vapor.

Transmittance

To do this, specific wavelengths of light, in this case from a tungsten-halogen lamp, are passed through the test solution.

Absorbance

Absorbance, Concentration relationship

Knowing the expected range of concentration of the desired species allows the sample to be prepared so that the Nova 60 can accurately measure the intensity of transmitted light. A limited number of Nova 60 test kits were available, so tests were limited to one membrane pressure for each of the solutions. An industrial magnetron charges the nitrogen plasma with microwave energy, which is then used to excite the atoms of the sample.

The conductivity of the permeate streams could not be measured using the available conductivity meters as the volume obtained during the run time was insufficient.

Solute Flux from Water Flux

Solute Flux

Solute Rejection

Solute Concentration in Permeate

Solute Mass Flowrate

Water Flux at constant TMP

Transmembrane Pressure

Water Recovery

Separation Factor

However, higher TMPs were chosen to ensure a sufficiently high flow rate so that the run time for the experiments was reasonable. This potential was determined by comparing the water flux through each of the membranes before and after the nanofiltration experiments. It can be seen that the degree of fouling increased with decreasing MWCO of the membranes.

For this reason, the separation of other ions relative to the phosphate ion was measured. The outcome of this project was twofold: first, to investigate the use of membrane systems in current and potential urine treatment processes as part of the Re-invent the Toilet Challenge; second, to identify and investigate the knowledge gaps required to implement such a membrane system, with a particular focus on the NF membrane stage. Fouling caused a negligible flux reduction for the NF 270 membrane, a 15% flux reduction for the NF 90 membrane, and an 18% reduction for the XLE membrane, which is within acceptable limits.